Comparison of three methods of microsatellitedetection

M. CHRISTENSEN, L. SUNDE, L. BOLUND & T. F. éRNTOFT

Departments of Clinical Biochemistry and Clinical Genetics, Aarhus University Hospital, Skejby,

Denmark

Christensen M, Sunde L, Bolund L, érntoft TF. Comparison of three methods

of microsatellite detection. Scand J Clin Lab Invest 1999; 59: 167±178.

Examination of microsatellites is frequent in the diagnosis of cancer.

Microsatellites are repeat DNA sequences scattered throughout the human

genome. These repeat regions are very frequent and highly polymorphic

elements. In this study we focus on dinucleotide repeats. We compared three

different methods for the detection of microsatellites: use of the ABI Prism 377

¯uorescence sequencer, autoradiography and silver-stained gels. DNA was

extracted from various clinical samples and ampli®ed by different polymerase

chain reaction (PCR) protocols. DNA from normal and tumor tissues was

analysed using each method. The ¯uorescence method was more sensitive than

the two other methods; however, this technology is very expensive. It seems

possible, when examining microsatellites on a low budget, to avoid radioactivity

by using silver-stained gels as an alternative. In conclusion, we observed identical

results when comparing autoradiography with the ¯uorescence technique.

However, we observed variability in the results when interpreting a single

locus comparing silver staining with autoradiography and the ¯uorescence

technique. Classi®cation of the tumors based on several microsatellite loci was

always identical.

Key words: ABI Prism 377; autoradiography; silver staining

Torben F. érntoft, Department of Clinical Biochemistry, Aarhus University Hospital,

Skejby 8200-Aarhus N, Denmark. Tel. z45 8949 5101, fax. z45 8949 6018,

e-mail. érntoft@kba.sks.aau.dk

INTRODUCTION

Three ``low-tech'' methods for visualizing

nucleic acids are available. The most commonly

used method is ethidium bromide staining in

agarose gels. This method is relatively fast, easy

and speci®c. However, ethidium bromide is a

powerful mutagen and its sensitivity is low. The

end product is not permanent. In addition,

ethidium bromide has a relatively low af®nity

for single-stranded DNA. Silver staining of

DNA in polyacrylamide gels and radioactive

labelling of nucleotides are very simple and

ef®cient methods of visualizing single-stranded

DNA. However, silver staining can be dif®cult

to control, and radioactive labelling requires

special facilities and is an expensive and time-

consuming process. Recently, tools for the

detection of ¯uorescence-labelled DNA (e.g.

the ABI Prism 377 DNA sequencer) have

Scand J Clin Lab Invest 1999; 59: 167 ± 178

167

Scan

d J

Clin

Lab

Inv

est D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y St

atsb

iblio

teke

t Tid

sskr

ifta

fdel

ing

on 0

3/09

/14

For

pers

onal

use

onl

y.

become available, but they require a rather

hefty equipment investment.

Optimizing methods for detection of micro-

satellite repeats are highly relevant, as micro-

satellite analysis is used to detect loss of

heterozygosity (LOH) and replication errors

(RER) in tumors. LOH is detected as reduced

intensity or total loss of one or more bands in

the tumor DNA compared with normal DNA

from the same individual. LOH has been

reported in several papers in association with

bladder cancer [1 ± 4]. RER is detected as

changes in the length of the microsatellite

sequences in the tumor DNA compared with

normal DNA [5 ± 7]. This instability is asso-

ciated with inactivation of the DNA mismatch

repair genes [8 ± 13]. We de®ne microsatellite

instability as new bands in the tumor DNA or

band shift with retained band pattern. The

replication error phenomenon is characteristic

in colorectal cancers developing in individuals

with HNPCC (Hereditary Non-Polyposis Colo-

rectal Cancer) [14 ± 16].

In the present study, we compared the

sensitivity of detecting PCR ampli®ed micro-

satellites by silver staining, autoradiography of33P-labelled products and ¯uorescence-labelled

products using the ABI 377 sequencer.

MATERIALS AND METHODS

In the ®rst part of the study we used DNA

extracted from colon cancer tissue and the

non-neoplastic resection border from the same

individual (silver staining/autoradiography).

The second part of the study (silver staining/

autoradiography/ABI Prism 377 sequencer) was

performed on DNA isolated from bladder

cancer tissue. In this case we used blood

DNA as a normal reference.

DNA extraction from tissue and blood

Instead of hazardous chloroform-phenol

DNA extraction, we isolated DNA using the

methods shown in Table I, which states the

sample sizes and yields with each method.

The extraction method used depended on the

available material.

High-salt DNA extraction from fresh or frozen

tissue

Materials were as follows. SLR buffer:

10 mM TRIS-HCl, pH 7.6, 5 mM MgCl26H2O,

10 mM NaCl; Lysis buffer: 10 mM TRIS-HCl,

pH 8.2, 2 mM EDTA Na2, pH 8.0, 400 mM NaCl;

TE buffer: 10 mM TRIS-HCl, pH 7.6, 1 mM

EDTA Na2, pH 8.0; 10% SDS (sodiumdode-

cylsulfate), 6 M NaCl, Proteinase K (1 mg/mL)

and 99% ethanol.

The tissue was homogenized and transferred

to a centrifuge tube containing 50 mL SLR

buffer. The tube was inverted several times and

centrifuged for 18 min at 2500 rpm, at 20³C.

The supernatant was discarded and the homo-

genate resuspended in 1 ± 2 mL SLR buffer.

Then 40 mL SLR buffer was added, the

suspension mixed, and centrifuged for 18 min

at 2500 rpm, 20³C. The supernatant was

discarded, and the homogenate resuspended in

4 mL Lysis buffer containing 200 mL 10% SDS

and 500 mL Proteinase K. The sample was

incubated with mild agitation in a water bath

for 1 h at 65³C, or incubated at 37³C overnight,

followed by cooling to room temperature (RT).

Then 1 mL 6 M NaCl was added and the

suspension vigorously mixed on a Vortex mixer

for 1.5 min and centrifuged at 2500 rpm for 18

min at 20³C, without the brake. The tube was

carefully removed from the centrifuge, and the

supernatant containing the DNA was poured

into a 50-mL centrifuge tube. To avoid

TABLE I. DNA extraction procedures, their material requirements and yield.

Method Tissue* or blood Yield (mg DNA)

High-salt extraction 50 ± 100 mg tissue 1.0 ± 20 mg/mg tissuePuregene 20 ± 50 mg tissue 0.5 ± 10 mg/mg tissueRapidPrep 5 ± 20 mg tissue 0.5 ± 10 mg/mg tissuePuregene 300 mL blood 15 ± 50 mg/mL blood

*Tumor tissue or tissue from the non-neoplastic resection border.

168 M. Christensen et al.

Scan

d J

Clin

Lab

Inv

est D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y St

atsb

iblio

teke

t Tid

sskr

ifta

fdel

ing

on 0

3/09

/14

For

pers

onal

use

onl

y.

contamination with salt and protein from the

pellet, 1 ± 2 mL of residual liquid was left. If a

pellet was not visible, the tube was centrifuged

once more. DNA in the supernatant was

precipitated by adding 30 mL 99% ethanol at

RT. The tube was inverted approx. 10 times

and left to rest for 15 min. Thereafter, it was

repeatedly inverted until the DNA threads

formed a visible precipitate. The DNA was

transferred with a micropipette to a 2-mL

Eppendorf tube and allowed to air dry for 1

h. Finally, the DNA was resuspended in 1 mL

TE buffer. The sample was left for 24 h at 57³Cbefore use, and stored at 4³C or ± 20³C.

Kit-based DNA extraction from fresh or frozen

tissue

The kit-based method used to extract DNA

depended on the sizes of the tissue samples

(Table I). All methods were carried out

according to the manufacturers' protocols. In

the kit provided by The Biotech Line (Pure-

Gene D-5000 DNA Isolation Kit, Minneapolis,

MN, USA), the method of DNA extraction

from tissue samples is basically the same as for

DNA extraction from blood samples (see

below).

Another kit for extracting DNA from tissue

was provided by Pharmacia Biotech, Inc.,

Sweden (RapidPrep Micro DNA Isolation

Kit). In short, the principle in this technique

is to take advantage of the highly anionic

nature of nucleic acids on an anion-exchange

chromatography column. The tissue was homo-

genized directly in a buffer containing guanidine

isothiocyanate, which disrupts cellular material

and inhibits nucleases. Following this treat-

ment, the sample was treated with RNase. The

sample was then added to the column, which

was washed twice. The DNA was eluted using a

high-salt alkaline buffer. Thereafter, the DNA

was precipitated with 100% isopropanol.

Finally, the DNA was resuspended in TE

buffer.

Kit-based DNA extraction from blood

Extraction of leukocyte DNA was carried out

with the kit provided by The Biotech Line

(PureGene D-5000 DNA Isolation Kit, Min-

neapolis, MN, USA). In short, anionic deter-

gents were added to the blood sample to lyse

the red and white blood cells. The lysate was

treated with RNase, and the proteins were

precipitated using a high-salt buffer solution.

Genomic DNA was precipitated with 100%

isopropanol and resuspended in a solution

containing a DNA preservative.

DNA extraction from formalin-®xed, paraf®n-

embedded tissue

DNA was extracted from 10 mm-thick sec-

tions from formalin-®xed, paraf®n-embedded

tissue. To obtain DNA from neoplastic and

non-neoplastic cells, sections were made from

blocks containing tumor tissue and tissue from

the resection border, respectively. DNA was

extracted as described by Madsen et al. [17],

with the modi®cation that the paraf®n was

removed by three incubations, each for 20 min,

with Vegeol (methyl esters from vegetable oils,

Aarhus Oliemoelle, Aarhus, Denmark) at 65³C.

PCR and electrophoresis conditions for silver

staining

The PCR for silver staining was conducted as

outlined in Table II. We used the following

microsatellite markers (provided by Research

Genetics, Inc., Huntsville, AL, USA):

D2S119 (2p) CTTGGGGAACAGAGGT-

CATT and GAGAATCCCTCAATTTCTTT-

GGA

D5S404 (5q) CTGGAGATGTAATGCTG-

TGC and GATCACCACATTCCACCTAAT

D8S255 (8p) TTTTGGAATTTCTAGCC-

TCC and TGAAACCCACAGATATTGGG

D10S197 (10p) ACCACTGCACTTCAGGT-

GAC and GTGATACTGTCCTCAGGTCTCC

D17S787 (17p) TGGGCTCAACTATATG-

AACC and TTGATACCTTTTTGAAGGGG

dNTP's (ATP, GTP, CTP, and TTP), Taq

DNA polymerase and PCR buffer were pro-

vided by Pharmacia Biotech. All the samples

were subjected to 35 cycles of PCR. The

annealing temperature was 58³C for all sets of

primers used. The ®nal extension was always

performed at 72³C for 5 min. Taq DNA

polymerase was added after the samples were

preheated to 95³C for 5 min (``hot start''). The

concentrations of DNA, nucleotides, primers

and Taq polymerase are listed in Table II. The

reaction volume was 20 mL. The PCR was

Comparison of three methods of microsatellite detection 169

Scan

d J

Clin

Lab

Inv

est D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y St

atsb

iblio

teke

t Tid

sskr

ifta

fdel

ing

on 0

3/09

/14

For

pers

onal

use

onl

y.

carried out in a Peltier Thermal cycler (PTC200,

MJ Research, MA, USA. Ten mL PCR

products were routinely evaluated by electro-

phoresis in high resolution 4% agarose gels,

FMC's NuSieve GTG, FMC BioProducts,

Rockland, ME, USA.

In addition to this were added 20 mL loading

buffer (95% formamide, 0.1 M NaOH, 0.05%

bromphenol blue, and 0.05% xylene cyanol),

after which the mixture was incubated at 95³Cfor 2 min, cooled on ice, and 6 mL were applied

to the gel. After having tested various poly-

acrylamide types and running conditions (see

below), we decided to perform the electrophor-

esis in a denaturing polyacrylamide gel: 1 X

MDE (Mutation Detection Enhancement) gel

(AT-Biotech, St Clara, CA, USA) containing

5.6 M urea and 32% formamide, using a 0.6 X

TBE running buffer. The temperature was kept

constant at 10³C by water cooling. The gel was

run at 4 W overnight. The gel was silver-stained

(see below) and a photo was taken (Eagle Eye

II, Stratagene, La Jolla, CA, USA). The photos

presented do not justify the results completely,

as the very ®ne threadlike bands are dif®cult to

reproduce, but are readily visible to the eye.

Optimizing gels for the silver staining procedure

We compared conventional and MDE poly-

acrylamide gels and tested various gel condi-

tions, as shown in Table III. The advantage of

MDE is that the gel is much less fragile than

conventional polyacrylamide gels, and thus

more resistant to the handling procedure

during silver staining. All gels contained urea

(5.6 M) and formamide (32%).

Condition A resulted in poor resolution with

dim bands and smiling effects. In condition B,

the resolution was better, but smiling effects

were still present. Smear was observed within

the band areas in condition C, but the cold tap

water prevented the smiling effects. The D and

E conditions resulted in the same resolution as

in F and G. The resolution in conditions F and

G was relatively good as regards the heavy

bands, but the light band areas were still

blurred. In the H condition, the resolution

was very ®ne as regards both the heavy and

light bands. Increasing the concentration of

MDE to 1.26MDE in condition I resulted in

poorer resolution compared with the 16MDE

gel. In future work we will use the 16MDE gel,

condition H, but for practical reasons we

increased the running time to overnight and

reduced the power to 4 W (condition J). The

temperature was still kept at 10³C. The

optimum amounts of 10% APS (ammonium

TABLE II. PCR conditions.

Silver staining Autoradiography

PCR buffer 1X 1XdNTP 150 mM each 100 mM eachPrimer I 5 pmol 2 pmolPrimer II 5 pmol 2 pmolTaq DNA

polymerase 0.20 U 0.15 UDNA 100 ng 30 ngReaction volume 20 mL 10 mL

TABLE III. Gel conditions.

Gel types Concentration Water cooling Power (W) Running time (h)

A MDE 0.56 No 6 4B MDE 0.86 No 6 4C MDE 0.56 cold tap water 6 4D Polyacr. 6% cold tap water 6 4E Polyacr. 9% cold tap water 6 4F MDE 0.56 RM 20 Lauda* 10 5G MDE 0.86 RM 20 Lauda 10 5H MDE 16 RM 20 Lauda 10 5I MDE 1.26 RM 20 Lauda 10 5J MDE 16 Julabo** 4 Overnight

MDE~MDE gel solution. Polyacr.~polyacrylamide. Acrylamide~bis, 19:1 (Bio Rad). Electrophoresisequipment: A ±E~Bio Rad (ProteanTM II), F ± J~Polar BearTM, S1SC. Power supply: A ±E~LBK BIO-CHROM 2103, F ± J~Consort E734. Plate dimensions: width6length (cm). A ±E~20624, F ± J~20640.Spacers (mm): A ±E~1, F ± J~0.4. Water cooling systems: *RM 20 Lauda, Lauda, Germany and **Julabo VC/123-Julabo FT200. The temperature was kept constant at 10³C.

170 M. Christensen et al.

Scan

d J

Clin

Lab

Inv

est D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y St

atsb

iblio

teke

t Tid

sskr

ifta

fdel

ing

on 0

3/09

/14

For

pers

onal

use

onl

y.

peroxide sulphate) and TEMED (tetramethyl-

ethylendiamine) were found to be 550 mL and

55 mL, respectively, in a total volume of 40 mL

gel solution. The glass plates were rinsed in

ethanol and wiped dry before casting the gel.

Silver staining

Silver staining was performed using the

following solutions:

A: 1% Nitric acid (HNO3);

B: 0.2% silver nitrate (AgNO3), 0.015%

formaldehyde;

C: 3% sodium carbonate (NaCO3), 0.015%

formaldehyde, 0.4% sodium thiosulphate

(Na2S2O3);

D: 10% acetic acid;

10% glycerol

The gel was soaked in solution A for 5 min and

washed in de-ionized water with three changes

of water, each for exactly 2 min. The gel was

then impregnated with silver nitrate by soaking

in solution B for approximately 30 min, with

mild agitation to prevent silver deposition on

the surface of the gel. This was followed by a

60-s rinse with de-ionized water. The gel was

developed in solution C ®rst by one quick dip.

Then the solution was changed once, and we

waited until the desired colour intensity of the

bands was achieved (2 ± 5 min; bands should

appear black against a light background).

Finally, the gel was soaked in solution D for

5 min and for another 5 min in solution E. The

gel was dried and stored by placing it between

transparent plastic sheets. It was kept between

absorbent paper until the next day and stored at

4³C.

It was found important to keep the tempera-

ture of solution C between 14³C and 16³C prior

to use. Likewise it was found that the wash

after soaking the gel in solution B should not

exceed 60 s. The solutions were used in 1.5-L

plastic polypropylene containers (5563566

cm). Other plastic materials seemed to absorb

the silver nitrate. After being removed from the

glass plates, the gel was handled on a ``gel lift''

constructed for this special purpose. The ``lift''

was a plastic tray (44630 cm) with two handles

and holes (1 mm) in the bottom around the

three edges (0.5 cm high). The holes made the

solution run off when changing from one

solution to another. The fourth side, which

had no edge, made it possible to transfer the gel

from one solution to another. This arrangement

made it quite easy to handle the relatively

fragile gels during the staining procedure.

PCR and electrophoresis conditions for

autoradiography

PCR for autoradiography was conducted

according to the procedures described in

Table II (PCR buffer: Boehringer-Mannheim,

Mannheim, Germany; primers: as described

above). One primer was end-labelled with 33P

using protein kinase (Gibco, Life Technologies,

Inc., Rockville, MD, USA). dNTPs were from

Life Technologies and Taq DNA polymerase

from Gibco. The concentrations of DNA,

nucleotides, primers and Taq polymerase are

listed in Table II. The PCRs were carried out in

microtiter plates. The reaction mixtures (10 mL)

were covered with 20 mL oil and incubated in a

PHC-3 Thermal Cycler (Techne, Cambridge,

England). The temperature pro®le for the PCR

was 94³C for 4 min, followed by 33 cycles of

85³C 1 s, 94³C 30 s, 55³C 30 s, 72³C 30 s, and a

®nal extension at 72³C for 10 min. The PCR

product was mixed with 10 mL of loading buffer

(95% formamide, 20 mM EDTA, 0.05% brom-

phenol blue and 0.05% xylene cyanol). The

mixture was incubated at 95³C for 10 min,

cooled on ice, and 2 mL were applied to the gel.

Electrophoresis was carried out in a 6%

polyacrylamide gel containing 7 M urea. A 1

X TBE running buffer was used. The gel

(width6length6thickness: 3063360.04 cm3)

was run for 1.5 ± 2 h at 60 W, giving a

temperature of approx. 60³C (BRL power

supply 2500, Gibco). The gel was transferred

to ®lter paper and dried for 2 h. Autoradio-

graphy was carried out overnight.

PCR conditions for analysis in the ABI Prism

377 sequencer

The forward primers of D2S119, D5S404,

D8S255 and D10S197 were 5' end-labelled with

FAM and HEX (¯uorescence amidites) using

standard techniques (Research Genetics, Inc.).

The PCR conditions were identical to those

used for silver staining. We made a mixture

containing 1 mL formamide, 0.5 mL loading

buffer (50 mg/mL blue dextran, 25 mM EDTA),

0.5 mL commercial standard (Tamra-500, ABI-

Comparison of three methods of microsatellite detection 171

Scan

d J

Clin

Lab

Inv

est D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y St

atsb

iblio

teke

t Tid

sskr

ifta

fdel

ing

on 0

3/09

/14

For

pers

onal

use

onl

y.

Perkin-Elmer, Norwalk, CT, USA) and 1.5 mL

diluted PCR product. The mixture was incu-

bated at 96³C for 4 min and cooled on ice. Then

1.5 mL of the mixture were applied to the gel

and electrophoresed in a 4.75% denaturing, 36-

cm polyacrylamide gel for 3 h, according to the

manufacturer's directions. Following electro-

phoresis, data were collected in the GeneScan

program for fragment analysis.

Scoring of gels

We scored the results according to micro-

satellite instability, band shift, or loss of

heterozygosity. Microsatellite instability was

identi®ed if (i) the tumor DNA gained

band(s) compared to normal DNA, (ii) if the

tumor DNA gained band(s) and lost bands, and

(iii) if the band pattern in the tumor DNA

shifted compared to normal. In order to be

scored as a shift, the band pattern had to be

retained. LOH was scored when a band in the

tumor was reduced or lost.

RESULTS

In the ®rst part of the study we analysed PCR

products from a patient with colon cancer in

microsatellite sequences in four different loci:

D5S404 D8S255, D10S197 and D17S787 in

silver-stained gels (Fig. 1A) and in gels exposed

to autoradiography (Fig. 1B). We extracted

tumor DNA from frozen colon tumor tissue,

and normal DNA was extracted from frozen

non-neoplastic tissue cut from the resection

border (see High salt DNA extraction from

frozen or fresh tissue). PCR conditions were as

given in the Methods section for silver-stained

gels and gels exposed to autoradiography.

The band patterns obtained with the two

methods were different. The autoradiography

method gave the ``classical'' stutter bands,

which were almost absent in the silver-stained

gels. However, both methods disclosed identical

patterns in the DNA from tumor and normal

tissue in each of the four examined loci.

Thereafter, we analysed PCR products in

three patients with colon cancer in a micro-

satellite marker located in locus D5S404 in a

silver-stained gel (Fig. 2A) and a gel exposed to

autoradiography (Fig. 2B). DNA was extracted

from formalin-®xed, paraf®n-embedded tissue

(see Materials and methods). Identical PCR

products were used for both types of gels. PCR

conditions were as given in the Methods section

for autoradiography. Moreover, we examined if

A

B

FIG. 1. Microsatellite analysis of four microsatellitemarkers, D5S404, D8S255, D10S197 and D17S787,in silver-stained gels and gels exposed to autoradio-graphy. Tumor DNA (T), normal DNA (N), con-trol (C), PCR without template. Corresponding setsof PCR products from tumor DNA (97) andnormal DNA (98) from a patient with colon cancerin a silver-stained gel (A) and a gel exposed toautoradiography (B). PCR conditions are given inthe Methods sections on silver-stained gels andautoradiography.

172 M. Christensen et al.

Scan

d J

Clin

Lab

Inv

est D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y St

atsb

iblio

teke

t Tid

sskr

ifta

fdel

ing

on 0

3/09

/14

For

pers

onal

use

onl

y.

the two methods were capable of detecting

relatively small concentrations of DNA. We

tested two different dilutions of DNA, 1 : 5 and

1 : 25, respectively.

The tumor DNA from patient 1 showed new

bands compared with normal DNA, whereas

the PCR products from tumor DNA and

normal DNA from patients 2 and 3 were

identical. There were no differences between the

two methods in ability to detect the diluted

DNA. Based on these data, we conclude that

the two methods have the same sensitivity with

respect to identifying microsatellite alterations.

Using the ¯uorescence technique has the

following advantages: it is possible to use very

low concentrations of PCR products, it gives

the exact length of a product because it has an

internal standard, and it supposedly minimizes

problems with stutter bands. We used this

technology in an ABI Prism 377 sequencer in

combination with silver-stained gels and gels

exposed to autoradiography. In a second study,

we examined DNA from ®ve bladder cancer

patients using the same four microsatellite

markers for each method: D2S119, D5S404,

D8S255 and D10S197. We extracted DNA from

A

B

FIG. 2. Microsatellite analysis of a microsatellitemarker, D5S404, in silver-stained gels and gelsexposed to autoradiography. Tumor DNA (T),normal DNA (N), control (C), PCR without tem-plate. (a) and (b) indicate that the DNA was diluted1 : 5 and 1 : 25, respectively. Corresponding sets ofidentical PCR products from tumor DNA andnormal DNA from three patients with colon cancerin a silver-stained gel (A) and a gel exposed toautoradiography (B). The PCR condition is des-cribed in the Methods section on autoradiography.

A B C

D E F

FIG. 3. Microsatellite analysis of two microsatellitemarkers, D2S119 (a ± c) and D8S255 (d ± f), in anABI Prism 377 sequencer, in gels exposed to auto-radiography, and silver-stained gels. Correspondingsets of tumor and normal DNA from two patientswith bladder cancer. ABI sequencer, (a), (d). Theallelic sizes (base pairs) are indicated on the tophorizontal axis of each panel in the electrophero-gram. Fluorescence units are on the vertical axis.The peak detection threshold was ®xed at 50 ¯uo-rescence units. The black solid line indicates tumorDNA and the white line represents normal DNA.Autoradiography, (b), (e). Silver-stained gels, (c), (f).Tumor DNA (T), normal DNA (N).

Comparison of three methods of microsatellite detection 173

Scan

d J

Clin

Lab

Inv

est D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y St

atsb

iblio

teke

t Tid

sskr

ifta

fdel

ing

on 0

3/09

/14

For

pers

onal

use

onl

y.

the tumor tissue and used blood from the same

individuals as normal DNA template (see Kit-

based DNA extraction from fresh or frozen

tissue or blood, respectively). PCR conditions

were as given in the Methods section for silver

staining and autoradiography. However, it was

not possible to detect the ¯uorescence-labelled

PCR products with the silver-stained gels.

Therefore, we made parallel PCRs, labelled

with 33P, ¯uorescence, or without labelling one

of the primers.

The ABI method proved much more sensitive

than silver staining and autoradiography, as the

PCR products had to be diluted approx. ten

times before loading on the ABI. Otherwise, the

signal was much too high.

When examining a single microsatellite

marker we did not always obtain identical

results with each of the three detection systems.

Fig. 3A, B and C shows divergent results. The

ABI sequencer (A) and autoradiography (B) are

identical, both showing instability. They are not

scored as LOH because of the instability. The

silver-stained gel (C), however, is scored as

LOH, as no instability is observed. Fig. 3D

(ABI sequencer), E (autoradiography), and F

(silver staining) show band shift with all three

methods. We found, when comparing auto-

radiography and the ABI method, autoradio-

graphy and silver staining, and silver staining

and the ABI method, 100% (16/16), 56.3% (9/

16) and 55.6.0% (10/18) concordance, respec-

tively. In 56.3% (9/16) of the cases, all three

methods gave identical results (Table IV). When

calculating the overall results, however, all three

methods classi®ed the ®ve tumors as highly

unstable, e.g. ¢ two unstable microsatellite loci

(Table V). This result is in accordance with a

previous study made in one of the laboratories

which included 4 bladder cancer patients, using

TABLE IV. Comparison of the results of the microsatellite analysis at identical microsatellite loci performedon an ABI sequencer, gels exposed to autoradiography and silver staining.

Marker and patient ABI sequencer Autoradiography Silver staining

D2S119Patient4 MIN/N (LOH) MIN/N (LOH)* LOH5 MIN/N (LOH) MIN/N (LOH)* LOH

15 MIN/S MIN/S MIN/N16 nc nc MIN/N20 MIN/N MIN/N MIN/N

D5S404Patient4 MIN/N MIN/N nc5 MIN/N MIN/N MIN/N

15 MIN/N MIN/N LOH16 MIN/S MIN/S nc20 MIN/S MIN/S MIN/S

D8S255Patient4 MIN/N nd MIN/N5 MIN/S nd MIN/N

15 MIN/N MIN/N MIN/S16 MIN/N MIN/N MIN/N20 MIN/S MIN/S MIN/S

D10S197Patient4 MIN/N MIN/N MIN/N5 MIN/N MIN/N MIN/N

15 LOH LOH LOH16 MIN/N MIN/N MIN/N20 nc nc MIN/N

MIN/N~microsatellite instability indicated as new bands in the tumor DNA compared to normal.MIN/S~ microsatellite instability as band shift with retained band pattern. MIN/N (LOH)~probably LOH,cannot be scored due to instability (we excluded these results, as it was not possible to give an exact score). *Weakstaining in the normal samples. nc~no change. nd~not determined.

174 M. Christensen et al.

Scan

d J

Clin

Lab

Inv

est D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y St

atsb

iblio

teke

t Tid

sskr

ifta

fdel

ing

on 0

3/09

/14

For

pers

onal

use

onl

y.

a panel of 22 different microsatellite markers. It

con®rms the correct classi®cation of the cases.

Patient 20 was not included [18].

DISCUSSION

The silver staining method appears to be as

sensitive as autoradiography in detecting the

ampli®ed dinucleotide fragments. Despite the

different patterns with silver-stained gels and

those exposed to autoradiography, the results

were identical in the ®rst part of the study. The

differences in band patterns were not unex-

pected, and are probably due to the denaturing

conditions and not least to the gel composition

in the two detection systems. In fact, the exact

composition of the MDE solution used for the

silver gels is not given, so it may be very

different from the gels used in the autoradio-

graphy.

We also compared the silver staining method,

autoradiography and fragment analysis per-

formed in the ABI Prism 377 sequencer using

the same microsatellite markers. The interpreta-

tion of a single microsatellite locus sometimes

gave divergent results. The concordance was

found to be 100% comparing autoradiography

and the ABI method. It was signi®cantly lower

comparing autoradiography and silver staining,

and silver staining and the ABI method. In

about 60% of the cases we obtained identical

results with all three methods. The complete

concordance between autoradiography and the

ABI method is probably due to the high

similarity in the composition of the gels. The

discrepancies in the interpretation of an indivi-

dual locus in silver-stained gels compared with

the other gel types tested are likely explained by

the plasticity of the silver-stained gels, but also

by the allele band patterns, which can be

dif®cult to interpret in silver-stained gels. It

seems, however, that discrepancies in the

interpretation of the individual microsatellites

do not in¯uence the overall classi®cation of

bladder tumors.

In a recent paper by Bocker et al., the same

conclusions were reached [19]. These authors

state the importance of studying a suf®cient

number of microsatellite loci to ensure the

correct classi®cation of the tumors. Based on

this, discrepancies at a single locus should not

change the overall result.

In the silver-stained gels we have found that

in some cases, ``stutter'' bands make interpreta-

tion dif®cult. They may complicate the decision

about which bands are real and which are just

PCR artefacts. Hauge & Litt [20] suggest that

slipped strand mispairing is the major mechan-

ism in generating these extra bands. In our

hands these bands were easier to interpret using

the ABI system, although they did not com-

pletely disappear.

An advantage of autoradiography is the

documentation based on the exposed ®lm,

which is storable for many years. With the

silver staining method, bands often become

faint after few months of storage, and taking a

photo, or making a photocopy, is necessary to

preserve the results. The results from the ABI

can be stored easily.

If only small samples, biopsies or tissue

sections are available, the ABI system is

preferable. The time required to perform the

tests does not seem to differ very much among

the methods.

The ABI method requires investment in

expensive equipment, whereas the other two

methods are based on conventional laboratory

equipment. The autoradiography method

requires radioactive isotopes that are expensive,

not to mention the hazards and security

TABLE V. Total instability in four microsatellite loci with each detection method.

TumourPatient ABI sequencer Autoradiography Silver staining classi®cation

4 3/3 2/2 2/3 H5 3/3 2/2 3/3 H

15 3/4 3/4 2/4 H16 3/4 3/4 3/4 H20 3/4 3/4 4/4 H

H~Highly unstable tumor.

Comparison of three methods of microsatellite detection 175

Scan

d J

Clin

Lab

Inv

est D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y St

atsb

iblio

teke

t Tid

sskr

ifta

fdel

ing

on 0

3/09

/14

For

pers

onal

use

onl

y.

requirements linked to autoradiography. With

that in mind, the silver staining method may be

attractive for microsatellite analysis in labora-

tories with small budgets.

In conclusion, despite the different band

patterns comparing DNA from colon tumors

and normal DNA in silver-stained gels and gels

exposed to autoradiography, no differences in

the results were obtained. Identical results were

also obtained comparing DNA from bladder

tumors with normal DNA using autoradio-

graphy and the ABI sequencer. There was

variability in the interpretation of individual

microsatellite loci when analysing normal DNA

and tumor DNA from patients with bladder

cancer using silver staining, autoradiography

and the ABI method. However, the overall

results were identical.

ACKNOWLEDGEMENTS

This research was supported by Aarhus Uni-

versity, Danish Cancer Research Fund, the

NOVO-Nordic Foundation and the Danish

Cancer Society. We thank Inge-Lis Thorsen,

Dept. of Clinical Biochemistry, Aarhus Uni-

versity Hospital, and Lisbeth Bùdker, Dept. of

Clinical Genetics, Aarhus University Hospital.

REFERENCES

1 Zulueta-Gonzalez M, Ruppert JM, Tokino K,Tsai YC, Spruck III CH, Miyao N, Nichols PW,Hermann GG, Horn T, Steven K, SummerhayesIC, Sidransky D, Jones PA. Microsatellite insta-bility in bladder cancer. Cancer Res 1993; 53:5620 ± 3.

2 Habuchi T, Devlin J, Elder PA, Knowles MA.Detailed deletion of chromosome 9q in bladdercancer: evidence for two tumour suppressor loci.Oncogene 1995; 11: 1671 ± 4.

3 Knowles MA, Shaw ME, Proctor AJ. Deletionmapping of chromosome 8 in cancers of theurinary bladder using restriction fragment lengthpolymorphisms and microsatellite polymorphisms.Oncogene 1993; 8: 1357 ± 64.

4 Shaw ME, Knowles MA. Deletion mapping ofchromosome 11 in carcinoma of the bladder.Genes Chromosomes Cancer 1995; 13: 1 ± 8.

5 Brasset C, Joyce JA, Froggatt NJ, Williams G,Furniss D, Walsh S, Miller R, Evans DG, MaherER. Microsatellite instability in early onset andfamilial colorectal cancer. J Med Genet 1996; 33:981 ± 5.

6 Liu B, Farrington SM, Petersen GM, HamiltonSR, Parsons R, Papadopoulos N, Fujiwara T, Jen

J, Kinzler KW, Wyllie AH, Vogelstein B, DunlopM. Genetic instability occurs in the majority ofyoung patients with colorectal cancer. NatureMedicine 1995; 1: 348 ± 52.

7 Lothe RA, Peltomaki P, Meling GI, Aaltonen LA,Nystrùm-Lahti M, Pylkkanen L, Heimdal K,Andersen TI, Mùller P, Rognum TO, FossaÊ SD,Haldorsen T, Langmark F, Brùgger A, de laChapella A, Bùrresen AL. Genomic instability incolorectal cancer: relationship to clinicopathologi-cal variables and family history. Cancer Res 1993;53: 5849 ± 52.

8 Dunlop MG, Farrington SM, Carothers AD,Wyllie AH, Sharp L, Burn J, Liu B, KinzlerKW, Vogelstein B. Cancer risk associated withgermline DNA mismatch repair gene mutations.Hum Mol Genet 1997; 6: 105 ± 10.

9 Moslein G, Tester DJ, Lindor NM, Honchel R,Cunningham JM, French AJ, Halling KC, SchwabM, Goretzki P, Thibodeau SN. Microsatelliteinstability and mutation analysis of hMSH2 andhMLH1 in patients with sporadic, familial andhereditary colorectal cancer. Hum Mol Genet1996; 5: 1245 ± 52.

10 Liu B, Parsons R, Papadopoulos N, NicolaidesNC, Lynch HT, Watson P, Jass JR, Dunlop M,Wyllie A, Peltomaki P, de la Chapelle A,Hamilton SR, Vogelstein B, Kinzler KW. Analysisof mismatch repair genes in hereditary non-polyposis colorectal cancer patients. Nature Med-icine 1996; 2: 169 ± 74.

11 Aaltonen LA, Peltomaki P. Genes involved inhereditary nonpolyposis colorectal carcinoma.Anticancer Res 1994; 14: 1657 ± 60.

12 Bronner EC, Baker SM, Morrison PT, Warren G,Smith LG, Lescoe MK, Kane M, Earabino C,Lipford J, Lindblom A, TannergaÊrd P, Bollag RJ,Godwin AR, Ward DC, Nordenskjold M, FischelR, Kolonder R, Liskay M. Mutation in the DNAmismatch repair gene homologue hMLH1 isassociated with hereditary non-polyposis cancer.Nature 1994; 368: 258 ± 61.

13 Nicolaides NC, Papadopoulos N, Liu B, Wei Y-F,Carter KC, Ruben SM, Rosen CA, Haseltine WA,Flieschmann RD, Fraser CM, Adams MD, VenterJC, Dunlop MG, Hamilton SR, Petersen GM, dela Chapelle A, Vogelstein B, Kinzler KW.Mutations of the two PMS homologues inhereditary nonpolyposis colon cancer. Nature1994; 371: 75 ± 80.

14 Young J, Leggett B, Gustafson C, Ward M, SearleJ, Thomas L, Buttenshaw R, Chenevix-Trench G.Genomic instability occurs in colorectal carcino-mas but not in adenomas. Hum Mutat 1993; 2:351 ± 4.

15 Aaltonen LA, Peltomaki P, Mecklin J-P, JarvinenH, Jass JR, Green JS, Lynch HT, Watson P,Tallqvist G, Juhola M, Sistonen P, Hamilton SR,Kinzler KW, Vogelstein B, de la Chapella A.Replication errors in benign and malignant tumorsfrom hereditary nonpolyposis colorectal cancerpatients. Cancer Res 1994; 54: 1645 ± 8.

16 Peltomaki P, Lothe RA, Aaltonen LA, PylkkanenL, Nystrùm-Lahti M, Seruca R, David L, Holm

176 M. Christensen et al.

Scan

d J

Clin

Lab

Inv

est D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y St

atsb

iblio

teke

t Tid

sskr

ifta

fdel

ing

on 0

3/09

/14

For

pers

onal

use

onl

y.

R, Rydberg D, Haugen A, Brùgger A, BùrresenAL, de la Chapella A. Microsatellite instability isassociated with tumors that characterize thehereditary non-polyposis colorectal carcinomasyndrome. Cancer Res 1993; 53: 5853 ± 5.

17 Madsen MK, Hasholt L, Berger J, Sùrensen SA.SSCP analysis of paraf®n vax embedded tissues ina family with an atypical form of Fabry disease.J Clin Pathol Mol Pathol 1996; 49: M310 ± 2.

18 Christensen M, Jensen MA, Wolf H, érntoft TF.Pronounced microsatellite instability in transi-tional cell carcinomas from young bladdercancer patients. Int J Cancer (Pred Oncol) 1998;79: 396 ± 401.

19 Bocker T, Diermann J, Friedl W, Gebert J,Holinski-Feder E, Karner-Hanusch J, vonKnebel-Doeberitz M, Koelbe K, Moeslein G,Schackert H-K, Wirtz H-C, Fishel R, RuschoffJ. Microsatellite instability analysis: a multicenterstudy for reliability and quality control. CancerRes 1997; 57: 4739 ± 43.

20 Hauge XY, Litt M. A study of the origin of``shadow bands'' seen when typing dinucleotiderepeat polymorphisms by the PCR. Hum MolGenet 1993; 2: 411 ± 5.

Received: 20 October 1998Accepted: 18 February 1999

Comparison of three methods of microsatellite detection 177

Scan

d J

Clin

Lab

Inv

est D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y St

atsb

iblio

teke

t Tid

sskr

ifta

fdel

ing

on 0

3/09

/14

For

pers

onal

use

onl

y.